Optimize artifact retrieval for runtimes without true subagents

examples/antigravity/sdd-orchestrator.md

@@ -41,15 +41,14 @@ SDD is the structured planning layer for substantial changes.
 - In `none`, do not write project files. Return results inline and recommend enabling `engram` or `openspec`.
 
 ### Commands
+
 - `/sdd-init` -> run `sdd-init`
 - `/sdd-explore <topic>` -> run `sdd-explore`
-- `/sdd-new <change>` -> run `sdd-explore` then `sdd-propose`
 - `/sdd-continue [change]` -> create next missing artifact in dependency chain
-- `/sdd-ff [change]` -> run `sdd-propose` -> `sdd-spec` -> `sdd-design` -> `sdd-tasks`
 - `/sdd-apply [change]` -> run `sdd-apply` in batches
 - `/sdd-verify [change]` -> run `sdd-verify`
 - `/sdd-archive [change]` -> run `sdd-archive`
-- `/sdd-new`, `/sdd-continue`, and `/sdd-ff` are meta-commands handled by YOU (the orchestrator). Do NOT invoke them as skills.
+- `/sdd-continue` are meta-command handled by YOU (the orchestrator). Do NOT invoke them as skills.
 
 ### Dependency Graph
 ```

skills/sdd-apply/SKILL.md

@@ -24,11 +24,22 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **CRITICAL: `mem_search` returns 300-char PREVIEWS, not full content. You MUST call `mem_get_observation(id)` for EVERY artifact. If you skip this, you will work with incomplete specs and produce wrong code.**
 
-  **STEP A — SEARCH** (get IDs only — content is truncated):
+  **STEP A — SEARCH** (ONLY for artifacts missing from prompt; get IDs only — content is truncated):
   1. `mem_search(query: "sdd/{change-name}/proposal", project: "{project}")` → save ID
   2. `mem_search(query: "sdd/{change-name}/spec", project: "{project}")` → save ID
   3. `mem_search(query: "sdd/{change-name}/design", project: "{project}")` → save ID

skills/sdd-archive/SKILL.md

@@ -23,11 +23,22 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **CRITICAL: `mem_search` returns 300-char PREVIEWS, not full content. You MUST call `mem_get_observation(id)` for EVERY artifact. If you skip this, you will archive with incomplete data.**
 
-  **STEP A — SEARCH** (get IDs only — content is truncated):
+  **STEP A — SEARCH** (ONLY for artifacts missing from prompt; get IDs only — content is truncated):
   1. `mem_search(query: "sdd/{change-name}/proposal", project: "{project}")` → save ID
   2. `mem_search(query: "sdd/{change-name}/spec", project: "{project}")` → save ID
   3. `mem_search(query: "sdd/{change-name}/design", project: "{project}")` → save ID

skills/sdd-design/SKILL.md

@@ -23,11 +23,22 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **CRITICAL: `mem_search` returns 300-char PREVIEWS, not full content. You MUST call `mem_get_observation(id)` for EVERY artifact. If you skip this, you will work with incomplete data and produce wrong design.**
 
-  **STEP A — SEARCH** (get IDs only — content is truncated):
+  **STEP A — SEARCH** (ONLY for artifacts missing from prompt; get IDs only — content is truncated):
   1. `mem_search(query: "sdd/{change-name}/proposal", project: "{project}")` → save ID
   2. `mem_search(query: "sdd/{change-name}/spec", project: "{project}")` → save ID (optional — may not exist if running in parallel with sdd-spec)
 

skills/sdd-explore/SKILL.md

@@ -23,8 +23,19 @@ The orchestrator will give you:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **Read context** (optional — load project context if available):
   1. `mem_search(query: "sdd-init/{project}", project: "{project}")` → get observation ID
   2. `mem_get_observation(id: {id from step 1})` → full project context
@@ -60,14 +71,15 @@ Read and follow `skills/_shared/persistence-contract.md` for mode resolution rul
 
 ### Retrieving Context
 
-Before starting, load any existing project context and specs per the active convention:
-- **engram**:
+Before starting, resolve context with prompt-first priority:
+- **prompt context first**: if the orchestrator already included project context/artifacts, use that directly.
+- **engram**: only if required context is missing from prompt:
   1. `mem_search(query: "sdd-init/{project}", project: "{project}")` → get observation ID
   2. `mem_get_observation(id: {id from step 1})` → full project context
   3. Optionally `mem_search(query: "sdd/", project: "{project}")` → find existing artifacts
   (If no results, proceed without prior context.)
-- **openspec**: Read `openspec/config.yaml` and `openspec/specs/`.
-- **none**: Use whatever context the orchestrator passed in the prompt.
+- **openspec**: if still missing, read `openspec/config.yaml` and `openspec/specs/`.
+- **none**: use whatever context the orchestrator passed in the prompt.
 
 ## What to Do
 

skills/sdd-propose/SKILL.md

@@ -24,9 +24,20 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
-  **Read dependencies** (two-step — search returns truncated previews):
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
+  **Read dependencies** (prompt-first: only use two-step retrieval for artifacts missing from prompt; search returns truncated previews):
   1. `mem_search(query: "sdd/{change-name}/explore", project: "{project}")` → get observation ID (optional — may not exist)
   2. If found: `mem_get_observation(id: {id})` → full exploration content
   3. `mem_search(query: "sdd-init/{project}", project: "{project}")` → project context (optional)

skills/sdd-spec/SKILL.md

@@ -23,11 +23,22 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **CRITICAL: `mem_search` returns 300-char PREVIEWS, not full content. You MUST call `mem_get_observation(id)` for EVERY artifact. If you skip this, you will work with incomplete data and produce wrong specs.**
 
-  **STEP A — SEARCH** (get IDs only — content is truncated):
+  **STEP A — SEARCH** (ONLY for artifacts missing from prompt; get IDs only — content is truncated):
   1. `mem_search(query: "sdd/{change-name}/proposal", project: "{project}")` → save ID
 
   **STEP B — RETRIEVE FULL CONTENT** (mandatory):

skills/sdd-tasks/SKILL.md

@@ -23,11 +23,22 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **CRITICAL: `mem_search` returns 300-char PREVIEWS, not full content. You MUST call `mem_get_observation(id)` for EVERY artifact. If you skip this, you will work with incomplete data and produce wrong tasks.**
 
-  **STEP A — SEARCH** (get IDs only — content is truncated):
+  **STEP A — SEARCH** (ONLY for artifacts missing from prompt; get IDs only — content is truncated):
   1. `mem_search(query: "sdd/{change-name}/proposal", project: "{project}")` → save ID
   2. `mem_search(query: "sdd/{change-name}/spec", project: "{project}")` → save ID
   3. `mem_search(query: "sdd/{change-name}/design", project: "{project}")` → save ID

skills/sdd-verify/SKILL.md

@@ -25,11 +25,22 @@ From the orchestrator:
 
 Read and follow `skills/_shared/persistence-contract.md` for mode resolution rules.
 
+Artifact resolution priority (prompt-first):
+1. Prompt context from orchestrator (highest priority)
+2. Engram persistence (`mem_search` -> `mem_get_observation`)
+3. Filesystem artifacts (`openspec`)
+
+If a required artifact is already present in the prompt context, use it directly and **DO NOT** query Engram for that artifact.
+Only query Engram for missing dependencies or when prompt content is explicitly partial/truncated.
+This keeps compatibility with runtimes that pass full artifacts inline or by reference.
+
 - If mode is `engram`:
 
+  Resolve dependencies using prompt-first priority. Query Engram only for artifacts not already provided in the prompt.
+
   **CRITICAL: `mem_search` returns 300-char PREVIEWS, not full content. You MUST call `mem_get_observation(id)` for EVERY artifact. If you skip this, you will verify against incomplete specs and miss issues.**
 
-  **STEP A — SEARCH** (get IDs only — content is truncated):
+  **STEP A — SEARCH** (ONLY for artifacts missing from prompt; get IDs only — content is truncated):
   1. `mem_search(query: "sdd/{change-name}/proposal", project: "{project}")` → save ID
   2. `mem_search(query: "sdd/{change-name}/spec", project: "{project}")` → save ID
   3. `mem_search(query: "sdd/{change-name}/design", project: "{project}")` → save ID